English text logic continuity analysis method

Abstract

The invention provides an English text logic coherence analysis method. The method is an analysis method composed of an English text logic relation tree generation module, an English text logic relation graph conversion module and an English text logic coherence analysis module which are connected in sequence. After an English text is processed through the analysis method, a logical continuity analysis result of the English text can be obtained. The problem that an existing English text coherence analysis method can only analyze the vocabulary coherence of the English text and cannot analyze the logic coherence of the English text is solved.

Description

Technical Field

[0001] The invention relates to natural language processing technology, and is a method for analyzing the logical coherence of an English text. The method of the invention is only suitable for analyzing the logical coherence quality of an English text, and is not suitable for analyzing the logical coherence quality of other texts. Background Art

[0002] The existing English text coherence analysis method extracts the coherence information of words and entities in the English text and uses machine learning and deep learning methods to judge the coherence degree of the English text. These existing English text coherence analysis methods can only analyze the lexical coherence of the English text, but cannot analyze the logical coherence of the English text. In view of the above problems, the present invention proposes an English text logical coherence analysis method, which analyzes the logical coherence quality of the English text based on the logical relationship of the English text. Summary of the invention

[0003] The present invention provides an English text logical coherence analysis method comprising: an English text logical relationship tree generation module, an English text logical relationship diagram conversion module, and an English text logical coherence analysis module. The overall processing flow chart is as follows: Figure 1 shown.

[0004] The processing flow of the English text logical relationship tree generation module of the present invention is: first, reading the English text to be analyzed, and performing sentence and word segmentation processing on the English text to be analyzed; second, performing logical relationship analysis on the English text after sentence and word segmentation, and constructing an English text logical relationship tree; third, outputting the constructed English text logical relationship tree.

[0005] The processing flow of the English text logical relationship diagram conversion module of the present invention is: first, read the English text logical relationship tree; second, traverse the English text logical relationship tree, and construct the edges and nodes of the English text logical relationship diagram according to the logical relationships in the English text logical relationship tree, wherein the edges are logical relationships, and the nodes connected by the edges are the core clauses of the first and second subtrees of the logical relationship, respectively. If there is no core clause, then the non-core clauses on the leftmost side of the first and second subtrees are selected as the core clauses; third, output the constructed English text logical relationship diagram.

[0006] The processing flow of the English text logical coherence analysis module of the present invention is: first, reading the English text logical relationship graph; second, according to the English text logical relationship graph, using the English text logical relationship graph adjacency grid calculation formula and the English text logical relationship graph feature grid calculation formula, respectively constructing the English text logical relationship graph adjacency grid and feature grid of the English text logical relationship graph, the English text logical relationship graph adjacency grid represents the connection relationship between nodes in the English text logical relationship graph, and the English text logical relationship graph feature grid represents the feature information of the nodes in the English text logical relationship graph; third, encoding the English text logical relationship graph adjacency grid and the English text logical relationship graph feature grid, using the English text logical relationship graph coherence vector calculation formula, to calculate the graph logical coherence vector of the English text; fourth, according to the English text graph logical coherence vector, using the English text logical coherence score calculation formula, calculate the logical coherence score of the English text, and output the logical coherence score and comments of the English text.

[0007] Concept and calculation formula of the present invention

[0008] (1) English text logical relationship graph adjacency grid calculation formula

[0009]

[0010] In formula (1), i and j represent the i-th node and j-th node in the English text logical relationship graph, respectively. The positions of the i-th node and the j-th node in the adjacency grid of the English text logical relationship graph are represented as i and j. When there is an edge between node i and node j, the corresponding value in the adjacency grid of the English text logical relationship graph is 1, otherwise it is 0.

[0011] (2) English text logical relationship diagram feature grid calculation formula

[0012]

[0013] In formula (2), n represents the total number of nodes in the English text logical relationship graph, w represents the word order information of the node, and m represents the total number of words in the node.

[0014] (3) Calculation formula of coherence vector in logical relationship diagram of English text

[0015]

[0016] In formula (3), i represents the i-th row corresponding to the i-th node in the English text logical relationship graph adjacency grid and the English text logical relationship graph feature grid, and n represents the total number of nodes in the English text logical relationship graph. The English text logical relationship graph adjacency grid is obtained by formula (1), and the English text logical relationship graph feature grid is obtained by formula (2)

[0017] (4) English text logical coherence score calculation formula

[0018]

[0019] In formula (4), j represents the jth number in the coherence vector of the English text logical relationship graph, m represents the total dimension of the coherence vector of the English text logical relationship graph, and e represents the base of the natural logarithm function. The coherence vector of the English text logical relationship graph is calculated by formula (3).

[0020] Specific processing steps of the correction method of the present invention

[0021] like Figure 1 As shown, the overall processing flow chart processing flow is as follows:

[0022] P101 starts;

[0023] P102 executes the English text logical relationship tree generation module. The specific steps are as follows Figure 2 As shown;

[0024] P103 executes the English text logical relationship diagram conversion module. The specific steps are as follows Figure 3 As shown;

[0025] P103 executes the English text logical coherence analysis module. The specific steps are as follows Figure 4 As shown;

[0026] P105 ends.

[0027] like Figure 2 As shown, the processing flow of the English text logical relationship tree generation module is as follows:

[0028] P201 starts;

[0029] P202 reads the English text to be analyzed;

[0030] P203 Sentence processing of the English text to be analyzed;

[0031] P204 performs word segmentation on the English text to be analyzed;

[0032] P205 analyzes the logical relationship of English text after sentence and word segmentation, and constructs the logical relationship tree of English text;

[0033] P206 outputs the constructed English text logical relationship tree;

[0034] P207 ends.

[0035] like Figure 3 As shown, the processing flow of the English text logical relationship diagram conversion module is as follows:

[0036] P301 starts;

[0037] P302 reads the logical relationship tree of English text;

[0038] P303 initializes the logical relationship queue;

[0039] P304 inputs the root node of the English text logical relationship tree into the logical relationship queue;

[0040] P305 dequeues the head node of the logical relationship queue and sets it as the current node;

[0041] P306 determines whether the root node of the first subtree of the current node is empty, if not, go to P307, if yes, go to P308;

[0042] P307 inputs the first subtree root node of the current node into the logical relationship queue;

[0043] P308 determines whether the root node of the second subtree of the current node is empty, if not, go to P309, if yes, go to P310;

[0044] P309 inputs the second subtree root node of the current node into the logical relationship queue;

[0045] P310 determines whether the current node is a leaf node, if not, go to P311, if yes, go to P313;

[0046] P311 sets the core clause of the first subtree of the current node to a node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause;

[0047] P312 sets the core clause of the second subtree of the current node to another node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause, and then the operation is transferred to P313;

[0048] P313 determines whether the logical relationship queue is empty, if not, go to P305 operation, if yes, go to P314 operation;

[0049] P314 outputs the logical relationship diagram of English text;

[0050] P315 ends.

[0051] like Figure 4 As shown, the processing flow of the English text logical coherence analysis module is as follows:

[0052] P401 starts;

[0053] P402 Reading English text logical relationship diagram;

[0054] P403 calculates the adjacency grid of the English text logical relationship graph according to the adjacency grid formula (1) of the English text logical relationship graph;

[0055] P404 represents the nodes of the logical relationship diagram of English text with feature vectors;

[0056] P405 calculates the characteristic grid of the English text logical relationship diagram according to the characteristic grid formula (2) of the English text logical relationship diagram;

[0057] P406 Calculate the coherence vector of the English text logic relationship diagram according to the coherence vector formula (3);

[0058] P407 Calculate the logical coherence score of the English text according to the English text logical coherence score formula (4);

[0059] P408 Generate comments on the logical coherence of English texts;

[0060] P409 ends.

[0061] like Figure 5 As shown, the logical relationship of the English text obtained based on the text to be analyzed is as follows:

[0062] The nodes in the graph represent clauses, and the edges between the nodes represent the relationship between two clauses. They are as follows:

[0063] Generalization relationship: Clause 1 and Clause 2, Clause 6 and Clause 8, Clause 8 and Clause 11, Clause 13 and Clause 18 are generalization relationships;

[0064] Elaboration relationship: Clause 2 and Clause 3, Clause 6 and Clause 7, Clause 11 and Clause 12, Clause 13 and Clause 14, Clause 14 and Clause 15, Clause 13 and Clause 17, Clause 6 and Clause 20, Clause 20 and Clause 21 are in an elaboration relationship;

[0065] Background relation: Clause 4 and Clause 5, and Clause 19 and Clause 20 are background relations;

[0066] Joint relation: Clause 2 and Clause 5, Clause 8 and Clause 13 are in joint relation;

[0067] Problem-solving relationship: Clause 1 and Clause 6 have a problem-solving relationship;

[0068] Method relationship: There is a method relationship between clause 8 and clause 9;

[0069] List relationship: Clause 9 and Clause 10 are in a list relationship;

[0070] Purpose relationship: There is a purpose relationship between Clause 16 and Clause 17. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 is the overall processing flow chart of the present invention;

[0072] Figure 2 It is a processing flow chart of the English text logical relationship tree generation module of the present invention;

[0073] Figure 3 It is a processing flow chart of the English text logical relationship diagram conversion module of the present invention;

[0074] Figure 4 is a processing flow chart of the English text logical coherence analysis module of the present invention;

[0075] Figure 5 It is the English text logical relationship diagram of the present invention. DETAILED DESCRIPTION

[0076] A method for analyzing logical coherence of an English text of the present invention comprises the following steps:

[0077] Step 1: Execute the "English text logical relationship tree generation module"

[0078] (1) The English text to be analyzed is as follows:

[0079] It is very important for college students to get to know the worldoutside the campus.For one thing,they can learn much that cannot be got frombooks from the world outside the campus.For another,as college students,getting to know the society will enable them to adapt to the society muchbetter after their graduation.There are several ways to get to know theworld,such as mediums and social services.First,college students can get toknow the world by reading newspaper,listening to radio and watchingtelevision.These mediums display a complex society before us.This is anindirect way.Second,we should go out of the campus and go into the society,take part in some practical social activity.For instance,we can go to theremote countryside to help the villagers develop education.Such practicalactivity and experience will make us have a further understanding about theworld.As for me,I will get to know the world outside the campus from mediums,and go into the society in my holidays.

[0080] (2) After the English text is segmented into sentences and words, the logical relationship tree of the English text is obtained as follows:

[0081]

[0082]

[0083] Step 2: Execute the "English text logical relationship diagram conversion module" and output the converted English text logical relationship diagram as follows: Figure 5 shown.

[0084] Step 3: Execute the "English text logical coherence analysis module"

[0085] (1) The adjacency grid of the logical relationship diagram of English text is as follows:

[0086] [[0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0087] [1,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0088] [0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0089] [0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0090] [0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0091] [1,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,0,1,0],

[0092] [0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0093] [0,0,0,0,0,1,0,0,1,0,1,0,1,0,0,0,0,0,0,0,0],

[0094] [0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0],

[0095] [0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0],

[0096] [0,0,0,0,0,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0],

[0097] [0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0],

[0098] [0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,1,1,0,0,0],

[0099] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0],

[0100] [0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0],

[0101] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0],

[0102] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0,0,0,0,0],

[0103] [0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0],

[0104] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0],

[0105] [0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1],

[0106] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]]

[0107] (2) The words in the English text to be analyzed are represented by vectors, and the resulting word vectors are as follows:

[0108] The first clause

[0109] [0.03328400105237961,-0.04075400158762932,...,-0.19683000445365906],[-0.17489999532699585,0.2295600026845932,...,0.36812999844551086],

[0110] [-0.19022999703884125,0.11810000240802765,...,0.10862000286579132],[0.237419992685318,0.10226999968290329,...,0.16607999801635742],

[0111] [-0.24131999909877777,0.12063000351190567,...,0.1550700068473816],

[0112] [-0.501479983329773,0.4590800106525421,...,0.22214999794960022],

[0113] [-0.4538100063800812,-0.007326600141823292,...,-0.1926800012588501],[-0.25756001472473145,-0.05713199824094772,...,-0.07062099874019623],[-0.14124000072479248,-0.1183599978685379,...,0.11568000167608261],[-0.25756001472473145,-0.05713199824094772,...,-0.07062099874019623],[-0.21053999662399292,0.13819999992847443,...,0.5273299813270569],

[0114] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],

[0115] [-0.2583099901676178,0.43643999099731445,...,-0.1410199999809265],

[0116] [-0.07607100158929825,0.34909000992774963,...,0.261029988527298],

[0117] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],[-0.16909000277519226,-0.37292999029159546,...,0.005296700168401003]

[0118] The second clause

[0119] [-0.24131999909877777,0.12063000351190567,...,0.1550700068473816],

[0120] [-0.3675599992275238,0.39500001072883606,...,0.0314910002052784],

[0121] [0.2646400034427643,-0.10502000153064728,...,0.4142700135707855],

[0122] [-0.26625001430511475,0.47964000701904297,...,-0.004142399877309799],[-0.5110200047492981,0.6175199747085571,...,-0.18386000394821167],

[0123] [-0.33156999945640564,-0.08879599720239639,...,0.001311299973167479],[-0.20117999613285065,-0.13655999302864075,...,-0.1729699969291687]

[0124] …

[0125] Clause 21

[0126] [0.03846599906682968,-0.039792001247406006,...,0.059702999889850616],[0.007367799989879131,0.06253200024366379,...,0.2936199903488159],

[0127] [-0.18233999609947205,-0.1920900046825409,...,0.176829993724823],

[0128] [0.046560000628232956,0.21318000555038452,...,0.05391300097107887],

[0129] [-0.10101000219583511,-0.3196200132369995,...,-0.15689000487327576],

[0130] [-0.44398999214172363,0.12816999852657318,...,-0.06255000084638596],

[0131] [-0.22745999693870544,-0.13658000528812408,...,0.5534600019454956],

[0132] [-0.009709199890494347,-0.31836000084877014,...,-0.4344399869441986]

[0133] (3) The characteristic grid of the logical relationship diagram of English text is as follows:

[0134] [[-2.568287007510662,1.726095444522798,...,1.0520546776242554],

[0135] [-1.6542600095272064,1.2824140042066574,...,0.24116991984192282],

[0136] [-1.7835206910967827,1.3384200185537338,...,-0.29846930457279086],

[0137] [-1.5155160017311573,1.0635534031316638,...,0.49550001323223114],

[0138] [-2.885171040892601,0.5408970136195421,...,-1.3463520868681371],

[0139] [-1.2205199748277664,0.5272980108857155,...,-0.3073039874434471],

[0140] [-2.831090033054352,0.37104197312146425,...,0.11796198971569538],

[0141] [-2.443009991198778,2.1822913689538836,...,0.4753919802606106],

[0142] [-0.6046049892902374,0.3198600113391876,...,-1.0426799952983856],

[0143] [0.24025099352002144,0.7467179894447327,...,0.31605999544262886],

[0144] [-1.4510132793802768,0.5167669774964452,...,-0.8604817083105445],

[0145] [-0.7387700080871582,0.7229420095682144,...,0.2152549959719181],

[0146] [-0.2675502044148743,0.7188690304756165,...,0.4160376996733248],

[0147] [-0.1909561986103654,-0.2757900133728981,...,0.42717598006129265],

[0148] [-1.1180119942873716,0.09529000520706177,...,-0.9624240007251501],

[0149] [-2.0874872291460633,1.272486001253128,...,0.29737259610556066],

[0150] [-1.61260100081563,0.2442139983177185,...,0.04682500660419464],

[0151] [-1.7310298702213913,1.0048769786953926,...,-0.19307767506688833],

[0152] [-0.302187598310411,0.10571999102830887,...,0.7520700246095657],

[0153] [-2.2252810150384903,0.7706560343503952,...,1.2530716476030648],

[0154] [-0.8721153875812888,-0.6025600209832191,...,0.4836459942162037]]

[0155] (4) The logical coherence analysis scores and comments of the English text are as follows:

[0156] According to formula (3) and formula (4), the logical coherence of the English text is analyzed, and the specific analysis results are as follows:

[0157] Logical coherence score: 0.673

[0158] Comments on logical coherence: The overall transition of the English text to be analyzed is basically natural, the logic is relatively clear, and the coherence is good.

Claims

1. A method for analyzing the logical coherence of an English text, characterized by: The method comprises a sequentially connected English text logical relationship tree generation module, an English text logical relationship diagram conversion module, and an English text logical coherence analysis module. The analysis method comprises the following processing flow: (1) The processing flow of the English text logical relationship tree generation module is as follows: first, read the English text to be analyzed, and perform sentence and word segmentation processing on the English text to be analyzed; second, perform logical relationship analysis on the English text after sentence and word segmentation, and construct the English text logical relationship tree; Third, output the constructed English text logical relationship tree; (2) The processing flow of the English text logical relationship diagram conversion module is: first, read the English text logical relationship tree; Second, traverse the English text logical relationship tree, and construct the edges and nodes of the English text logical relationship graph according to the logical relationships in the English text logical relationship tree. The edges are logical relationships, and the nodes connected by the edges are the core clauses of the first and second subtrees of the logical relationship. If there is no core clause, then the leftmost non-core clauses of the first and second subtrees are selected as the core clauses; Third, output the constructed English text logical relationship graph; (3) The processing flow of the English text logical coherence analysis module is as follows: first, read the English text logical relationship diagram; second, according to the English text logical relationship diagram, use the English text logical relationship diagram adjacency grid calculation formula and the English text logical relationship diagram feature grid calculation formula to respectively construct the English text logical relationship diagram adjacency grid and feature grid of the English text logical relationship diagram, the English text logical relationship diagram adjacency grid represents the connection relationship between nodes in the English text logical relationship diagram, and the English text logical relationship diagram feature grid represents the feature information of nodes in the English text logical relationship diagram; Third, the adjacency grid of the English text logic relationship graph and the feature grid of the English text logic relationship graph are encoded, and the graph logic coherence vector of the English text is calculated using the English text logic relationship graph coherence vector calculation formula; Fourth, according to the logical coherence vector of the English text graph, the logical coherence score of the English text is calculated using the English text logical coherence score calculation formula, and the logical coherence score and comments of the English text are output.

2. The method for analyzing the logical coherence of an English text according to claim 1, characterized in that: The processing steps of the English text logical relationship tree generation module are as follows: P201 starts; P202 reads the English text to be analyzed; P203 Sentence processing of the English text to be analyzed; P204 performs word segmentation on the English text to be analyzed; P205 analyzes the logical relationship of English text after sentence and word segmentation, and constructs the logical relationship tree of English text; P206 outputs the constructed English text logical relationship tree; P207 ends.

3. The method for analyzing the logical coherence of an English text according to claim 1, characterized in that: The processing steps of the English text logical relationship diagram conversion module are as follows: P301 starts; P302 reads the logical relationship tree of English text; P303 initializes the logical relationship queue; P304 inputs the root node of the English text logical relationship tree into the logical relationship queue; P305 dequeues the head node of the logical relationship queue and sets it as the current node; P306 determines whether the root node of the first subtree of the current node is empty, if not, go to P307, if yes, go to P308; P307 inputs the first subtree root node of the current node into the logical relationship queue; P308 determines whether the root node of the second subtree of the current node is empty, if not, go to P309, if yes, go to P310; P309 inputs the second subtree root node of the current node into the logical relationship queue; P310 determines whether the current node is a leaf node, if not, go to P311, if yes, go to P313; P311 sets the core clause of the first subtree of the current node to a node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause; P312 sets the core clause of the second subtree of the current node to another node of the current edge. If there is no core clause, the leftmost non-core clause is selected as the core clause, and then the operation is transferred to P313; P313 determines whether the logical relationship queue is empty, if not, go to P305 operation, if yes, go to P314 operation; P314 outputs the logical relationship diagram of English text; P315 ends.

4. The method for analyzing the logical coherence of an English text according to claim 1, characterized in that: The processing steps of the English text logical coherence analysis module are as follows: P401 starts; P402 Reading English text logical relationship diagram; P403 calculates the adjacency grid of the English text logical relationship graph according to the adjacency grid formula (1) of the English text logical relationship graph; P404 represents the nodes of the logical relationship diagram of English text with feature vectors; P405 calculates the characteristic grid of the English text logical relationship diagram according to the characteristic grid formula (2) of the English text logical relationship diagram; P406 Calculate the coherence vector of the English text logic relationship diagram according to the coherence vector formula (3); P407 Calculate the logical coherence score of the English text according to the English text logical coherence score formula (4); P408 Generate comments on the logical coherence of English texts; P409 ends.

5. The method for analyzing logical coherence of English text according to claim 1, characterized in that: The calculation formula of the English text logical coherence analysis module is defined as follows: (1) English text logical relationship graph adjacency grid calculation formula In formula (1), i and j represent the i-th node and j-th node in the English text logical relationship graph, respectively. The positions of the i-th node and the j-th node in the adjacency grid of the English text logical relationship graph are represented by i and j. When there is an edge between node i and node j, the corresponding value in the adjacency grid of the English text logical relationship graph is 1, otherwise it is 0; (2) English text logical relationship diagram feature grid calculation formula In formula (2), n represents the total number of nodes in the English text logical relationship graph, w represents the word order information of the node, and m represents the total number of words in the node; (3) Calculation formula of coherence vector in logical relationship diagram of English text In formula (3), i represents the i-th row corresponding to the i-th node in the adjacency grid of the English text logical relationship graph and the feature grid of the English text logical relationship graph, n represents the total number of nodes in the English text logical relationship graph, the adjacency grid of the English text logical relationship graph is obtained by formula (1), and the feature grid of the English text logical relationship graph is obtained by formula (2); (4) English text logical coherence score calculation formula In formula (4), j represents the jth number in the coherence vector of the English text logical relationship graph, m represents the total dimension of the coherence vector of the English text logical relationship graph, and e represents the base of the natural logarithm function. The coherence vector of the English text logical relationship graph is calculated by formula (3).

Images